Add runtime provenance to benchmark metadata

[Genmin]

Summary

• add a shared runtime provenance helper for Python/package versions and committed dependency lockfile hashes
• include runtime provenance in reference-output metadata and scenario-manifest metadata
• include generated timestamp and runtime provenance in no-tools resume metadata

Fixes #6

Validation

• uv run --extra dev pytest -q (189 passed)
• uv run --extra dev ruff check policybench/provenance.py policybench/eval_no_tools.py policybench/cli.py tests/test_eval_no_tools.py tests/test_provenance.py
• uv run --extra dev ruff format --check policybench/provenance.py policybench/eval_no_tools.py policybench/cli.py tests/test_eval_no_tools.py tests/test_provenance.py
• git diff --check

[vercel]

@Genmin is attempting to deploy a commit to the PolicyEngine Team on Vercel.

A member of the Team first needs to authorize it.